The International Committee on Taxonomy of Viruses (ICTV) has recently approved a proposal to create a plant virus family Virgaviridae. The family is named because its members have rod-shaped virions (from the Latin virga = rod), and it includes the genera Furovirus, Hordeivirus, Pecluvirus, Pomovirus, Tobamovirus and Tobravirus. The chief characteristics of members of the family are:

  1. 1.

    Alpha-like replication proteins that form a distinct phylogenetic “family” [5].

  2. 2.

    Single-stranded RNA + sense genomes with a 3′-t-RNA-like structure and no polyA tail.

  3. 3.

    Rod-shaped virions 20–25 nm in diameter with a central “canal”.

  4. 4.

    Coat proteins of 19–24 kDa.

It contains some viruses in which there is a single cell-to-cell movement protein (MP) of the ‘30K’ superfamily [7] and others that encode a triple gene block (TGB) [8]. There are also differences in the number of genomic RNAs (1, 2 or 3 depending on the genus), but sequence analysis of the polymerase and other genes suggests that the viruses form a coherent taxonomic unit (see below). Some properties of the six genera included in the family are summarized in Table 1, and their genome organization is shown in Fig. 1. Biologically, the viruses are fairly diverse. They have been reported from a wide range of herbaceous and mono- and dicotyledonous plant species, but the host range of individual members is usually limited. All members can be transmitted experimentally by mechanical inoculation, and for those in the genus Tobamovirus, this is the only known means of transmission. In some genera, transmission is by soil-borne vectors, while members of the genus Hordeivirus are transmitted through pollen and seed. The only genus with rod-shaped virions excluded from this list is Benyvirus, because this is much more distantly related in phylogenetic analyses of the polymerase (see below) and because (unlike the others) its members have a polyadenylated genome and a polymerase that is processed by autocatalytic protease activity.

Table 1 Major properties of the genera included in the new family Virgaviridae
Full size table
Fig. 1
figure 1

Diagram showing the genome organization of the six genera included in the family Virgaviridae. Domains marked in the replication proteins are Methyltransferase (M), Helicase (H) and RNA-dependent RNA polymerase (R). Triple gene block proteins (TGB) are cross-hatched, and coat proteins are in black. MP, movement protein of the ‘30K’ superfamily; C, cysteine-rich protein. Positions of “leaky” stop codons are shown by triangles (filled triangles). tVal/Tyr/His/−: t-RNA-like structure accepting valine, tyrosine, histidine or not aminoacylated, respectively. Brackets indicate ORFs that are missing from some strains

Full size image

On the basis of their analysis of the RNA-dependent RNA polymerase (RdRp) gene from a wide range of viruses, Koonin and Dolja [5] included viruses from the six genera described in this paper within RdRp Supergroup 3, which they sub-divided into three lineages that they suggested might correspond to orders. One of these lineages, which they named Tobamo, included the six genera considered here, together with the families Closteroviridae and Bromoviridae and the genus Idaeovirus. Phylogenetic analysis (using several different methods) of the RdRp domain, of the whole replication protein or of the fused Methyl transferase–Helicase–RdRp domains continues to support this grouping and shows that the genus Benyvirus is much too distantly related to be grouped in this family (see Fig. 2). The inclusion within the branch of the families Closteroviridae and Bromoviridae also justifies the inclusion of all six genera within the single family Virgaviridae. The replication proteins constitute the majority of the genomes of these viruses and provide the best phylogenetic trees, but there are also indications of relatedness amongst the other genes. For example, the TGB proteins of the genera Hordeivirus, Pecluvirus and Pomovirus are clearly related and form a distinct group separate from those of the genus Benyvirus and the filamentous viruses in the family Flexiviridae (recently split into two families). A tree for TGBp1 sequences is provided in Fig. 3, and more details supporting this classification of the TGB proteins are provided by Morozov and Solovyev [8]. The small size of the coat protein and its inherent variability make it less suitable for phylogenetic analysis. Nevertheless, significant groupings of genera occur (Furo- with Pomo-; Peclu- with Hordei- and Tobra- a bit more distant) which correspond with those found within the RdRp (Fig. 4). There are also close relationships between the small cysteine-rich proteins of Furovirus, Hordeivirus, Pecluvirus and Tobravirus, although those of Pomovirus do not align well with them (data not shown).

Fig. 2
figure 2

Phylogenetic tree of the amino acid sequences of the fused Met–Hel–RdRp domains of the members of the six genera included in the family Virgaviridae together with some other related viruses. Distantly related genera and families that formed well-supported monophyletic clades were collapsed into a triangle, the length of which corresponds to the variation found within the clade. The recently established order Tymovirales includes the families Tymoviridae and Flexiviridae (which has also been divided). The neighbour-joining (NJ) tree is shown, but nearly identical trees were produced from the alignment using Maximum Composite Likelihood (ML) and Bayesian tree building algorithms. Percentage bootstrap support (out of 1,000 replications) for NJ and ML trees and posterior probability for the Bayesian tree are, respectively, indicated on the corresponding branches separated by slashes if they differed from each other. Values are only indicated on the major branches when >60%, and when values were identical, only one number is indicated (asterisk). The consensus tree generated by ML did not support the inclusion of BBNV into a Pomovirus clade and grouped the genus Idaeovirus within the Bromoviridae clade. The scale indicates JTT amino acid distances. Alignments were made from translated nucleotide sequences using the ClustalW algorithm in the Alignment Explorer module of MEGA4 [9] as described previously [6]. A total of 500 amino acid positions corresponding to 1,500 nt positions were used for the alignment. NJ and ML trees were generated using standard settings for these algorithms in MEGA4 [9] from protein and back-translated nucleotide alignments, respectively. The Bayesian tree was generated from back-translated nucleotide alignment using MrBayes v3.1.2 [4], employing the general time reversible model with gamma-shaped rate variation with a proportion of invariable sites; 1,000,000 generations of MCMC analysis were the point at which the average standard generation of split frequency between two parallel runs had reached 0.009565

Full size image
Fig. 3
figure 3

Phylogenetic (neighbor-joining) tree of the amino acid sequences of the TGBp1 proteins of members of the genera included in the family Virgaviridae together with other TGB-containing viruses. Numbers on branches indicate percentage of bootstrap support out of 1,000 bootstrap replications (when >60%). The scale indicates JTT amino acid distances. Tree produced in MEGA4 [9]. A tree of similar typology was obtained by maximum-likelihood analysis (PROML in PHYLIP [3])

Full size image
Fig. 4
figure 4

Phylogenetic (neighbor-joining) tree of the amino acid sequences of the coat proteins of members of the genera included in the family Virgaviridae. Numbers on major branches indicate percentage of bootstrap support out of 1,000 bootstrap replications (when >60%). The scale indicates JTT amino acid distances. Tree produced in MEGA4 [9]. A tree of similar typology was obtained by maximum-likelihood analysis (PROML in PHYLIP [3])

Full size image

The taxonomic structure of the new family and the species currently included are listed in Table 2.

Table 2 List of species recognised within the genera belonging to the new family Virgaviridae with accession numbers for complete genome nucleotide sequences
Full size table

Sequence differences between and within the existing species in the family were examined and compared with molecular criteria for species discrimination provided by the relevant study groups in the 8th ICTV report [2]. Individual pairwise comparisons were therefore made using the nt and aa sequences of each fully sequenced gene from every available accession in the family Virgaviridae contained in the international databases. Comparisons used the GCG [1] program GAP (with a gap creation penalty of 50 and a gap extension penalty of 3 for nt comparisons and values of 8 and 2, respectively, for aa comparisons). This program aligns the two sequences selected and calculates the percentage identity and similarity between them. To assist with the large numbers of comparisons, software was written (Antoniw, unpublished) to generate batch files that were run in GCG and also to extract and summarize data from the output files. Some of the chief features of the data for the replication protein, the RdRp and the coat protein genes are summarized in Table 3. Within some genera, there are rather few species and sequences, but some conclusions may nevertheless be reached. For the genus Tobravirus, it is already known that coat protein sequences (from RNA2) are of little taxonomic value [2], and this appears also to be the case for the genus Pecluvirus. Within the replication protein and RdRp, isolates of the same species usually had >90% nt or aa sequence identity. Comparisons between genera show that some existing species in the genus Tobamovirus are rather closely related and there may be merit in re-examining the species demarcation criteria within this genus.

Table 3 Values from pairwise sequence comparisons for three genome regions amongst viruses in the family Virgaviridae
Full size table